---

title: “Time series analysis project” output: html_notebook

1. A brief background of the data set, including the source and any background information on the type of variable collected.

I am looking at the stock price data of a Fortune 500 company named Owens Corning. The data was collected by Yahoo Finance through which I managed to scrape the data into csv file format. There are several variables to look at in this dataset that include opening, closing, high, low and adj. closing stock price. My goal is to look at the stock price over several years instead of any particular day. Hence I choose the weekly adj. closing price as my time series data which represents the closing stock price after adjusting for the actions taken by the company. The data is collected every week along with the additional data for when dividends were released (I ignore this data to ensure equally spaced time periods; Some data cleansing steps were performed to exclude this data).

2. Constructing the time-series plot of the data.

OC_stock <- read.csv("https://docs.google.com/spreadsheets/d/1pI9XgBe1TuPmqdvSzkLofA9lJxb0iIODOsX_-MfeHGQ/pub?output=csv")
temp=OC_stock[OC_stock$Dividend==0,]
temp$Date=as.Date(temp$Date, "%m-%d-%Y")
plot(temp$Date,temp$Adj.Close, ylab="Adjusted Closing Stock Price",xlab="Time", type="l", pch=16)

3. Observations made from the time-series plot:

1] Trend: The first notable observation is that although there’s fluctuation in the data over time, the general trend seems to be increasing (or upward). This means that the closing stock value of the company has been generally increasing since 2008.

2] There’s an all time low closing price of about $4 which is around the end of the 2008 financial crisis, and a high of about $54 recently in 2016.

3] Cycle: As of any long term cycle, there does not appear to be any obvious pattern. There appears to be some short term cycles every year but they are not obvious as well.

4] Seasonality: If we ignore the data before 2010 and after 2015, there appear to be a seasonality every year. The closing stock value rises roughly around the first quarter with a local high around second/third quarter, and falls during the fourth quarter. This pattern is again difficult to distinguish from the noise.

5] Irregular behavior: There’s a lot of irregular behavior in the data especially around 2008-10 during when the US was recovering from the financial crisis which may be the reason for it. The closing stock value stars soaring rapidly ending 2015 which indicates some economic factors into play. Naturally there’s a lot of external factors affecting the stock prices, hence forecasting may not be very accurate.